Update of janusplot from 0.1.0 to 0.1.1. The current CRAN version is 0.1.0
(published 2026-04-28). This is a minor feature and performance release with no
breaking change to the public API.
- New
engine = c("bam", "gam")argument selecting themgcvfitting backend. The default changes frommgcv::gamtomgcv::bamfor a 3-10x speed-up at janusplot's scale. This is the single non-byte-identical change:bam's fREML estimation differs fromgam's REML by ~1-3% in effective degrees of freedom, so fitted values, the asymmetry index, and per-cell colour fills may shift slightly when a user upgrades.engine = "gam"reproduces 0.1.0 output verbatim; this is documented prominently at the top ofNEWS.md. - Per-cell k-checking, scale-aware compact rendering, rendering-only axes modes,
and figure-to-file output via
save_as. All strictly additive. - Internal: shared scalar-argument validation centralised behind one helper; test coverage of argument validation extended.
- macOS 26 (Darwin 25) / Apple Silicon -- R 4.5.2 (local).
- win-builder (R-devel and R-release) -- to be re-run immediately before submission.
- R-hub v2 (
rhub::rhub_check(), default platforms) -- to be re-run immediately before submission.
Local R CMD check --as-cran on the built janusplot_0.1.1.tar.gz: 0 ERRORs,
0 WARNINGs, 2 NOTEs. Both NOTEs are local-environmental and are not expected on
the CRAN check farm, win-builder, or R-hub.
checking for future file timestamps ... unable to verify current time is
raised on the local check host, which cannot reach a time service.
checking HTML version of manual ... Skipping checking HTML validation: 'tidy' doesn't look like recent enough HTML Tidy reflects the dated tidy shipped with
the local macOS host; the CRAN farm and win-builder carry a recent HTML Tidy and
do not raise it.
There are 0 reverse dependencies on CRAN (checked against the current CRAN package database on 2026-06-26: zero reverse Depends / Imports / LinkingTo, and zero reverse Suggests). The default-engine change therefore has no reverse- dependency impact.
janusplot renders pairwise asymmetric smoothed-association matrices of
continuous variables via mgcv GAM fits. Each off-diagonal cell displays a
directional smooth; the package exposes per-pair shape descriptors
(monotonicity, convexity, inflection counts, a 24-category shape taxonomy) plus
a recovery-rate sensitivity study (janusplot_shape_sensitivity()) with
precomputed demo data.
- All examples are executable without
\dontrun{}; the heavier demonstrations sit under\donttest{}. - All tests run under
testthatedition 3; visual output is regression-tested withvdiffrand skips cleanly where the snapshot back-end is unavailable. - No compiled code; pure R with
Imports:frommgcv,ggplot2,patchwork,grid,stats,cli,lifecycle,rlang. - Source tarball is ~2 MB, including the precomputed
shape_sensitivity_demodataset.